utaufandomcom-20200223-history
Help:Voice recording checklist
Settings and adjustments Voicebank recording checklist In preparation, make sure to have the following prior to recording a personal UTAU voicebank. *'Location' Very important. *'Equipment' Equally important is the equipment. Typically will use either a desktop, laptop or a netbook. *'Microphone' However little, this equipment CAN make or break recordings. *'Headset' Before finalizing the sample, it must undergo test, which includes listening to the samples before saving. *'Editing tools' To cut out excess blanks, this will be needed. *'Voice Source' What good is the recording without one? Kana vs Romaji recording There are several ways to record a bank. The most popular among overseas users tend to be recording in either Audacity or OREMO. Both give you an option to name in either hiragana or romaji. The choice is up to the voicers on which encoding to use. Romaji Pros: *Easier for non-Japanese users to read. *Usable regardless of locale. Cons: *Adding hiragana aliases is a must, as many USTs are made in hiragana. *Romaji VCV has few premade oto.ini's, resulting in users having to create one from scratch. Kana Pros: *Helps with learning hiragana. *Automatically compatible with hiragana usts. *VCV oto.ini's can be generated using setparam. Cons: *Cannot be used by users without Japanese locale. *May confuse some users, resulting in mispronunciations. Other language compatibility It is not recommended to use kana to express sounds from other languages. If one is planning on making a bank in almost any other language, they are required to use latin characters. Since UTAU can't process file names written in non-kana or non-latin languages (e.g. Korean), people must write their recordlists (or save their files) in latin characters. It's not a good idea to try and express sounds from another language in kana. Post-recording tweaking After recording, editing the samples is an option to make a better voicebank. *'Pitch' If some samples are off-pitch, pitch correction software can be used to apply the correct pitch, such as Autotune, Melodyne, and NewTone. *'Dynamics' If some samples are quieter or louder than others, you can edit the volume using virtually any music editing software, such as Audacity, Cubase, or FL Studio. *'Equalization' For added characteristic vocals, EQ effects can be applied with virtually any music software. Voicebank compensation When a voicebank is badly recorded, there are a few things you can do to fix it. This applies to voicebanks that are your own. *'Add flags' in order to improve the voice and add desirable/good-sounding effects. A quick fix for some samples that just don't sound quite right. *'Check the oto.ini.' If it is missing or badly done, try redoing it in order to fix problems. *'Apply a filter' to all samples. Modern sound editing softwares have capabilities of batch application of filters and effects. This should remedy problems that extend to several samples, where individual editing and cleanup is considered time-consuming. Extra Visualization and aesthetics An UTAU's voice must have a certain amount of two things in order to be an aesthetically sound voicebank. *'Depth' An UTAU's voice must be recording in a way as to invoke a feeling of the vocal you are trying to capture. Depth is a quality that makes a voice more or less vivid. Triphone/VCV voicebanks usually have much more depth to them the Diphone/CV voicebanks. *'Characteristic vocals' Your UTAU most by the embodiment of your voice, so don't try to mimic another UTAU's voice unless the UTAU is a product of voice acting. Record a voicebank with confidence in order to make an UTAU unique. Alias encoding Alias Encoding allows your UTAU to understand two languages at once. First, open up UTAU, and make sure you have all of the samples you need. Then click (T) then (S). It should open up all of your samples. Find a sample you want to alias. For example find 'a', click it, then second box from the top, paste 'あ'. Now click OK. Now when UTAU runs hiragana UST's, 'a' will be scanned as hiragana and romaji.